[mythtv-users] Problems running mythtv-setup on backend with Intel 82865G

[Tom Dexter]

On Thu, Jul 9, 2009 at 6:30 PM, Michael T. Dean<mtdean at thirdcontact.com> wrote:
> On 07/09/2009 05:23 PM, Tom Dexter wrote:
>>
>> On Thu, Jul 9, 2009 at 4:37 PM, Tom Dexter wrote:
>>
>>>
>>> On Thu, Jul 9, 2009 at 4:06 PM, Tom Dexter wrote:
>>>
>>>>
>>>> I've been reading nightmarish accounts of problems getting DRI working
>>>> with the 865G chipsets under xorg 1.5.3, however I've never even used
>>>> DRI on that machine before and still don't have it in my kernel at
>>>> all.  What's really confusing me is that glxinfo says "Direct
>>>> Rendering: yes" even though no drm modules are even loaded...that
>>>> really seems off.  I've tried everything I can find to force disable
>>>> DRI in xorg.conf, but glxinfo _still_ says direct rendering is
>>>> enabled...any idea what that's all about?  I have no idea if that has
>>>> anything to do with is anyway.
>>>
>>> Wow...what do you know.  I was on the right track with the above.
>>>
>>> If I do:
>>>
>>> export LIBGL_ALWAYS_INDIRECT=1
>>>
>>> ...before startx the problem goes away and mythtv-setup runs just fine.
>>>
>>>
>>
>> I've also noticed that with that environment variable set on the
>> backend, all the ssh/X11Forwarding situations I described in the first
>> post are able to run mythtv-setup just fine.  That leads me to believe
>> that this may be partly/mainly a bug in the X11 libraries running on
>> the backend, as none of those situations involve the backend
>> xorg-server or it's Intel video.  Ugly stuff for sure.
>
> I thought it was drivers.  It's also required for ATI drivers.  I'm 99%
> positive it's not required for NVIDIA drivers.
>
> Mike

Yea...when run via ssh from the two machines I have running nVidia
drivers it worked fine.

This whole mess appears to be a mesa bug:

http://bugs.gentoo.org/show_bug.cgi?id=277099

I just installed that mesa patch and that did it.  Just to confuse
things, all that erroneous stuff with direct rendering showing as
enabled when running the backends X server had nothing whatsoever to
do with the crash, and has probably been the case all along.

...a painful one for sure.

Tom